Nine years ago Mr. Markey sued his former employer Tenneco under Title VII, bringing an individual claim of discriminatory discharge and class claims of discriminatory hiring, discharge, promotion and pay practices regarding blacks in entry level positions. On a prior appeal, we affirmed the district court's dismissal of his individual claim and the class claims of discriminatory discharge, promotion and pay, as well as that court's decision to narrow the class to all black applicants to Tenneco's entry level bargaining unit from March 15, 1974, to the date of the final order. Markey v. Tenneco Oil Company, 635 F.2d 497 (5th Cir.1981). We remanded, however, for further findings as to the relevant labor market and a determination, based on that market, whether Tenneco had engaged in a pattern or practice of discrimination in hiring.

At the first trial, Markey had contended that the relevant labor market for the Tenneco refinery in question was the New Orleans Standard Metropolitan Statistical Area (SMSA) in which, according to the 1970 census, the percentage of unskilled black laborers was 59.7%. The district court rejected this approach, choosing instead to weight the population of black laborers in each of the four parishes in the SMSA by the residence of Tenneco employees when hired. It did so, in part at least, because of its finding that the refinery was not accessible by public transportation and that therefore dispersion of blacks in the SMSA did not correlate with the labor pool actually available to Tenneco. We determined, however, that it was improper to weight the black population in each parish by the number of Tenneco employees residing there because this would allow an employer to determine the racial composition of his relevant market by restricting his hires to particular areas. 635 F.2d at 500-01. So holding, we suggested:

We recognize and share the district court's concern that the SMSA may not accurately reflect the labor market actually available to Tenneco. The SMSA approach, for example, may give equal weight to areas far from the plant in which significant numbers of workers may never even contemplate working for Tenneco. We cannot tell from the record whether the SMSA encompasses geographical areas from which job applicants are unlikely to come. We think it would be within the trial court's discretion on remand to modify its approach and assign a statistical weight to the percentage of blacks in each parish based on that parish's contribution to the applicant pool. Absent discriminatory recruiting practices, the percentage of applicants from a particular parish may be probative of the willingness of individuals in that parish to travel to the Tenneco plant and of the relative accessibility of the plant to residents of the parish, and thus be a more accurate measure of that parish's contribution to Tenneco's labor pool. The trial court should, of course, consider any evidence that helps define the areas from which Tenneco would normally be expected to draw its employees.

Id. (emphasis added).

On remand, the district court adopted our suggestion. Working from additional evidence submitted, and weighting the black population of each parish by the percentage of Tenneco applicants coming from it during the relevant times, the court derived a theoretical labor market from which about 42 percent of Tenneco's expectable applicants should have been black. Since the evidence also showed, without significant dispute, that almost 40 percent of the actual applicants and about half of the hires during this period had been black, the court concluded that no pattern or practice of discrimination was shown.1 Markey again seeks reversal, this time on a narrow ground: that because Tenneco's recruiting practices during the period in question were discriminatory, the applicant flow model adopted on remand by the district court was an improper one.

His burden is a heavy one. The determination of a relevant labor market is one of fact, reviewable only for clear error. Pullman-Standard v. Swint, 456 U.S. 273, 102 S. Ct. 1781, 72 L. Ed. 2d 66 (1982); see Castaneda v. Pickard, 648 F.2d 989, 1003 (5th Cir.1981) (views issue as "essentially factual" even before Swint ). Both in Castaneda, however, and on the earlier appeal of this case, we cautioned against uncritical acceptance of applicant flow as a basis of determining labor market in a case where that flow might be skewed by discriminatory recruiting practices. Markey now contends that this is such a case, pointing to the location of the refinery in predominantly white St. Bernard Parish and to opinion testimony that word of mouth recruiting among the refinery's eighty percent white work force located there would be likely to replicate that racial composition. His argument finds some support in our authorities, e.g., United States v. Georgia Power Co., 474 F.2d 906 (5th Cir.1973) (93 percent white work force). Common sense indicates that such a result might be expected.

Many a promising theory founders on the facts, however,2 and those evident in our record demonstrate that word of mouth recruiting at this refinery has not proved a discriminatory practice. Since it has not, the caveat that we voiced on our earlier remand is satisfied, the court's fact-finding on the relevant labor market is not clearly erroneous, and Markey's appeal must fail.

To begin with, Mr. Markey's showing of statistical disparity in black hires is, at most, borderline. Even accepting Mr. Markey's original claim that the SMSA figure of just under 60 percent expectable black hires was the proper one for comparison, as well as that its relation to the actual black hires of about 50 percent during the relevant period is statistically significant in the abstract, its practical significance seems doubtful in view of the plant's location in 96 percent white St. Bernard Parish, a circumstance that no one contends is the result of any discriminatory animus, and of the want of public transportation to the refinery discerned by the trial court--two neutral conditions that, in combination, seem in reason to promote a higher percentage of whites in the work force than might otherwise be expected. Perhaps in consequence, Tenneco, the record reveals, has been assiduous in pursuing hiring practices either racially neutral or, in some respects, slanted toward the black applicant. At the top of every employment application found in the record appears the following legend:

NOTICE: This information will not be used for the purpose of discrimination. Any item on this form which you feel tends to be discriminatory need not be completed. Employment and advancement in Tenneco and its Associated Companies is determined by a person's qualifications and abilities without regard to race, color, age, religion, sex, national origin, physical & mental handicapped or veterans. It is our policy to treat each individual who applies for work and those subsequently hired in a fair and equitable manner.

That is not quite true, however, since the record also shows and the trial court found that during the relevant time period Tenneco pursued a "one-for-one" black/white hiring policy, one that necessarily resulted in the hiring of a larger percentage of its 38 percent black applicant flow than of its white applicants. Whatever the merits or faults of such a policy, it is scarcely one of which Mr. Markey or the class can complain.

But more to the point at issue, the prediction that word-of-mouth recruiting in the work force would be likely to replicate its composition did not come true. Though only one-fifth of the general work force was black during the period, almost two-fifths of the applicants were so. And in addition, the record seems to indicate that half of those hired during the period knew no one at all in the refinery work force and that, of those who did, almost half were black.3 We cannot say, in view of such evidence, that the district court's adoption of a labor market based on applicant flow, an approach suggested by us in our former opinion, was clearly erroneous. Since this is so, appellant's parallel contention that the court should have adopted his proposed model, one based upon transportation time zones, must likewise fail.

AFFIRMED.